net/octeontx2: support Rx

Add Rx burst support.

Signed-off-by: Jerin Jacob <jerinj@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Harman Kalra <hkalra@marvell.com>

drivers/net/octeontx2/Makefile

@@ -31,6 +31,7 @@ LIBABIVER := 1
 # all source are stored in SRCS-y
 #
 SRCS-$(CONFIG_RTE_LIBRTE_OCTEONTX2_PMD) += \
+	otx2_rx.c 	\
 	otx2_tm.c	\
 	otx2_rss.c	\
 	otx2_mac.c	\

drivers/net/octeontx2/meson.build

@@ -2,7 +2,7 @@
 # Copyright(C) 2019 Marvell International Ltd.
 #
 
-sources = files(
+sources = files('otx2_rx.c',
 		'otx2_tm.c',
 		'otx2_rss.c',
 		'otx2_mac.c',

drivers/net/octeontx2/otx2_ethdev.c

@@ -14,12 +14,6 @@
 
 #include "otx2_ethdev.h"
 
-static inline void
-otx2_eth_set_rx_function(struct rte_eth_dev *eth_dev)
-{
-	RTE_SET_USED(eth_dev);
-}
-
 static inline void
 otx2_eth_set_tx_function(struct rte_eth_dev *eth_dev)
 {

drivers/net/octeontx2/otx2_ethdev.h

@@ -280,6 +280,7 @@ struct otx2_eth_dev {
 	struct otx2_eth_qconf *tx_qconf;
 	struct otx2_eth_qconf *rx_qconf;
 	struct rte_eth_dev *eth_dev;
+	eth_rx_burst_t rx_pkt_burst_no_offload;
 	/* PTP counters */
 	bool ptp_en;
 	struct otx2_timesync_info tstamp;
@@ -482,6 +483,7 @@ int otx2_ethdev_parse_devargs(struct rte_devargs *devargs,
 			      struct otx2_eth_dev *dev);
 
 /* Rx and Tx routines */
+void otx2_eth_set_rx_function(struct rte_eth_dev *eth_dev);
 void otx2_nix_form_default_desc(struct otx2_eth_txq *txq);
 
 /* Timesync - PTP routines */

drivers/net/octeontx2/otx2_rx.c

@@ -0,0 +1,129 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(C) 2019 Marvell International Ltd.
+ */
+
+#include <rte_vect.h>
+
+#include "otx2_ethdev.h"
+#include "otx2_rx.h"
+
+#define NIX_DESCS_PER_LOOP	4
+#define CQE_CAST(x)		((struct nix_cqe_hdr_s *)(x))
+#define CQE_SZ(x)		((x) * NIX_CQ_ENTRY_SZ)
+
+static inline uint16_t
+nix_rx_nb_pkts(struct otx2_eth_rxq *rxq, const uint64_t wdata,
+	       const uint16_t pkts, const uint32_t qmask)
+{
+	uint32_t available = rxq->available;
+
+	/* Update the available count if cached value is not enough */
+	if (unlikely(available < pkts)) {
+		uint64_t reg, head, tail;
+
+		/* Use LDADDA version to avoid reorder */
+		reg = otx2_atomic64_add_sync(wdata, rxq->cq_status);
+		/* CQ_OP_STATUS operation error */
+		if (reg & BIT_ULL(CQ_OP_STAT_OP_ERR) ||
+		    reg & BIT_ULL(CQ_OP_STAT_CQ_ERR))
+			return 0;
+
+		tail = reg & 0xFFFFF;
+		head = (reg >> 20) & 0xFFFFF;
+		if (tail < head)
+			available = tail - head + qmask + 1;
+		else
+			available = tail - head;
+
+		rxq->available = available;
+	}
+
+	return RTE_MIN(pkts, available);
+}
+
+static __rte_always_inline uint16_t
+nix_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts,
+	      uint16_t pkts, const uint16_t flags)
+{
+	struct otx2_eth_rxq *rxq = rx_queue;
+	const uint64_t mbuf_init = rxq->mbuf_initializer;
+	const void *lookup_mem = rxq->lookup_mem;
+	const uint64_t data_off = rxq->data_off;
+	const uintptr_t desc = rxq->desc;
+	const uint64_t wdata = rxq->wdata;
+	const uint32_t qmask = rxq->qmask;
+	uint16_t packets = 0, nb_pkts;
+	uint32_t head = rxq->head;
+	struct nix_cqe_hdr_s *cq;
+	struct rte_mbuf *mbuf;
+
+	nb_pkts = nix_rx_nb_pkts(rxq, wdata, pkts, qmask);
+
+	while (packets < nb_pkts) {
+		/* Prefetch N desc ahead */
+		rte_prefetch_non_temporal((void *)(desc + (CQE_SZ(head + 2))));
+		cq = (struct nix_cqe_hdr_s *)(desc + CQE_SZ(head));
+
+		mbuf = nix_get_mbuf_from_cqe(cq, data_off);
+
+		otx2_nix_cqe_to_mbuf(cq, cq->tag, mbuf, lookup_mem, mbuf_init,
+				     flags);
+		otx2_nix_mbuf_to_tstamp(mbuf, rxq->tstamp, flags);
+		rx_pkts[packets++] = mbuf;
+		otx2_prefetch_store_keep(mbuf);
+		head++;
+		head &= qmask;
+	}
+
+	rxq->head = head;
+	rxq->available -= nb_pkts;
+
+	/* Free all the CQs that we've processed */
+	otx2_write64((wdata | nb_pkts), rxq->cq_door);
+
+	return nb_pkts;
+}
+
+
+#define R(name, f5, f4, f3, f2, f1, f0, flags)				\
+static uint16_t __rte_noinline	__hot					       \
+otx2_nix_recv_pkts_ ## name(void *rx_queue,				       \
+			struct rte_mbuf **rx_pkts, uint16_t pkts)	       \
+{									       \
+	return nix_recv_pkts(rx_queue, rx_pkts, pkts, (flags));		       \
+}									       \
+
+NIX_RX_FASTPATH_MODES
+#undef R
+
+static inline void
+pick_rx_func(struct rte_eth_dev *eth_dev,
+	     const eth_rx_burst_t rx_burst[2][2][2][2][2][2])
+{
+	struct otx2_eth_dev *dev = otx2_eth_pmd_priv(eth_dev);
+
+	/* [TSTMP] [MARK] [VLAN] [CKSUM] [PTYPE] [RSS] */
+	eth_dev->rx_pkt_burst = rx_burst
+		[!!(dev->rx_offload_flags & NIX_RX_OFFLOAD_TSTAMP_F)]
+		[!!(dev->rx_offload_flags & NIX_RX_OFFLOAD_MARK_UPDATE_F)]
+		[!!(dev->rx_offload_flags & NIX_RX_OFFLOAD_VLAN_STRIP_F)]
+		[!!(dev->rx_offload_flags & NIX_RX_OFFLOAD_CHECKSUM_F)]
+		[!!(dev->rx_offload_flags & NIX_RX_OFFLOAD_PTYPE_F)]
+		[!!(dev->rx_offload_flags & NIX_RX_OFFLOAD_RSS_F)];
+}
+
+void
+otx2_eth_set_rx_function(struct rte_eth_dev *eth_dev)
+{
+	const eth_rx_burst_t nix_eth_rx_burst[2][2][2][2][2][2] = {
+#define R(name, f5, f4, f3, f2, f1, f0, flags)				\
+	[f5][f4][f3][f2][f1][f0] =  otx2_nix_recv_pkts_ ## name,
+
+NIX_RX_FASTPATH_MODES
+#undef R
+	};
+
+	pick_rx_func(eth_dev, nix_eth_rx_burst);
+
+	rte_mb();
+}

drivers/net/octeontx2/otx2_rx.h

@@ -15,7 +15,10 @@
 					 PTYPE_TUNNEL_ARRAY_SZ) *\
 					 sizeof(uint16_t))
 
+#define NIX_RX_OFFLOAD_NONE            (0)
+#define NIX_RX_OFFLOAD_RSS_F           BIT(0)
 #define NIX_RX_OFFLOAD_PTYPE_F         BIT(1)
+#define NIX_RX_OFFLOAD_CHECKSUM_F      BIT(2)
 #define NIX_RX_OFFLOAD_VLAN_STRIP_F    BIT(3)
 #define NIX_RX_OFFLOAD_MARK_UPDATE_F   BIT(4)
 #define NIX_RX_OFFLOAD_TSTAMP_F        BIT(5)
@@ -30,4 +33,248 @@ struct otx2_timesync_info {
 	uint8_t		rx_ready;
 } __rte_cache_aligned;
 
+union mbuf_initializer {
+	struct {
+		uint16_t data_off;
+		uint16_t refcnt;
+		uint16_t nb_segs;
+		uint16_t port;
+	} fields;
+	uint64_t value;
+};
+
+static __rte_always_inline void
+otx2_nix_mbuf_to_tstamp(struct rte_mbuf *mbuf,
+			struct otx2_timesync_info *tstamp, const uint16_t flag)
+{
+	if ((flag & NIX_RX_OFFLOAD_TSTAMP_F) &&
+	    mbuf->packet_type == RTE_PTYPE_L2_ETHER_TIMESYNC &&
+	    (mbuf->data_off == RTE_PKTMBUF_HEADROOM +
+	     NIX_TIMESYNC_RX_OFFSET)) {
+		uint64_t *tstamp_ptr;
+
+		/* Deal with rx timestamp */
+		tstamp_ptr = rte_pktmbuf_mtod_offset(mbuf, uint64_t *,
+						     -NIX_TIMESYNC_RX_OFFSET);
+		mbuf->timestamp = rte_be_to_cpu_64(*tstamp_ptr);
+		tstamp->rx_tstamp = mbuf->timestamp;
+		tstamp->rx_ready = 1;
+		mbuf->ol_flags |= PKT_RX_IEEE1588_PTP | PKT_RX_IEEE1588_TMST
+			| PKT_RX_TIMESTAMP;
+	}
+}
+
+static __rte_always_inline uint64_t
+nix_clear_data_off(uint64_t oldval)
+{
+	union mbuf_initializer mbuf_init = { .value = oldval };
+
+	mbuf_init.fields.data_off = 0;
+	return mbuf_init.value;
+}
+
+static __rte_always_inline struct rte_mbuf *
+nix_get_mbuf_from_cqe(void *cq, const uint64_t data_off)
+{
+	rte_iova_t buff;
+
+	/* Skip CQE, NIX_RX_PARSE_S and SG HDR(9 DWORDs) and peek buff addr */
+	buff = *((rte_iova_t *)((uint64_t *)cq + 9));
+	return (struct rte_mbuf *)(buff - data_off);
+}
+
+
+static __rte_always_inline uint32_t
+nix_ptype_get(const void * const lookup_mem, const uint64_t in)
+{
+	const uint16_t * const ptype = lookup_mem;
+	const uint16_t lg_lf_le = (in & 0xFFF000000000000) >> 48;
+	const uint16_t tu_l2 = ptype[(in & 0x000FFF000000000) >> 36];
+	const uint16_t il4_tu = ptype[PTYPE_NON_TUNNEL_ARRAY_SZ + lg_lf_le];
+
+	return (il4_tu << PTYPE_WIDTH) | tu_l2;
+}
+
+static __rte_always_inline uint32_t
+nix_rx_olflags_get(const void * const lookup_mem, const uint64_t in)
+{
+	const uint32_t * const ol_flags = (const uint32_t *)
+			((const uint8_t *)lookup_mem + PTYPE_ARRAY_SZ);
+
+	return ol_flags[(in & 0xfff00000) >> 20];
+}
+
+static inline uint64_t
+nix_update_match_id(const uint16_t match_id, uint64_t ol_flags,
+		    struct rte_mbuf *mbuf)
+{
+	/* There is no separate bit to check match_id
+	 * is valid or not? and no flag to identify it is an
+	 * RTE_FLOW_ACTION_TYPE_FLAG vs RTE_FLOW_ACTION_TYPE_MARK
+	 * action. The former case addressed through 0 being invalid
+	 * value and inc/dec match_id pair when MARK is activated.
+	 * The later case addressed through defining
+	 * OTX2_FLOW_MARK_DEFAULT as value for
+	 * RTE_FLOW_ACTION_TYPE_MARK.
+	 * This would translate to not use
+	 * OTX2_FLOW_ACTION_FLAG_DEFAULT - 1 and
+	 * OTX2_FLOW_ACTION_FLAG_DEFAULT for match_id.
+	 * i.e valid mark_id's are from
+	 * 0 to OTX2_FLOW_ACTION_FLAG_DEFAULT - 2
+	 */
+	if (likely(match_id)) {
+		ol_flags |= PKT_RX_FDIR;
+		if (match_id != OTX2_FLOW_ACTION_FLAG_DEFAULT) {
+			ol_flags |= PKT_RX_FDIR_ID;
+			mbuf->hash.fdir.hi = match_id - 1;
+		}
+	}
+
+	return ol_flags;
+}
+
+static __rte_always_inline void
+otx2_nix_cqe_to_mbuf(const struct nix_cqe_hdr_s *cq, const uint32_t tag,
+		     struct rte_mbuf *mbuf, const void *lookup_mem,
+		     const uint64_t val, const uint16_t flag)
+{
+	const struct nix_rx_parse_s *rx =
+		 (const struct nix_rx_parse_s *)((const uint64_t *)cq + 1);
+	const uint64_t w1 = *(const uint64_t *)rx;
+	const uint16_t len = rx->pkt_lenm1 + 1;
+	uint64_t ol_flags = 0;
+
+	/* Mark mempool obj as "get" as it is alloc'ed by NIX */
+	__mempool_check_cookies(mbuf->pool, (void **)&mbuf, 1, 1);
+
+	if (flag & NIX_RX_OFFLOAD_PTYPE_F)
+		mbuf->packet_type = nix_ptype_get(lookup_mem, w1);
+	else
+		mbuf->packet_type = 0;
+
+	if (flag & NIX_RX_OFFLOAD_RSS_F) {
+		mbuf->hash.rss = tag;
+		ol_flags |= PKT_RX_RSS_HASH;
+	}
+
+	if (flag & NIX_RX_OFFLOAD_CHECKSUM_F)
+		ol_flags |= nix_rx_olflags_get(lookup_mem, w1);
+
+	if (flag & NIX_RX_OFFLOAD_VLAN_STRIP_F) {
+		if (rx->vtag0_gone) {
+			ol_flags |= PKT_RX_VLAN | PKT_RX_VLAN_STRIPPED;
+			mbuf->vlan_tci = rx->vtag0_tci;
+		}
+		if (rx->vtag1_gone) {
+			ol_flags |= PKT_RX_QINQ | PKT_RX_QINQ_STRIPPED;
+			mbuf->vlan_tci_outer = rx->vtag1_tci;
+		}
+	}
+
+	if (flag & NIX_RX_OFFLOAD_MARK_UPDATE_F)
+		ol_flags = nix_update_match_id(rx->match_id, ol_flags, mbuf);
+
+	mbuf->ol_flags = ol_flags;
+	*(uint64_t *)(&mbuf->rearm_data) = val;
+	mbuf->pkt_len = len;
+
+	mbuf->data_len = len;
+}
+
+#define CKSUM_F NIX_RX_OFFLOAD_CHECKSUM_F
+#define PTYPE_F NIX_RX_OFFLOAD_PTYPE_F
+#define RSS_F	NIX_RX_OFFLOAD_RSS_F
+#define RX_VLAN_F  NIX_RX_OFFLOAD_VLAN_STRIP_F
+#define MARK_F  NIX_RX_OFFLOAD_MARK_UPDATE_F
+#define TS_F	NIX_RX_OFFLOAD_TSTAMP_F
+
+/* [TSMP] [MARK] [VLAN] [CKSUM] [PTYPE] [RSS] */
+#define NIX_RX_FASTPATH_MODES						       \
+R(no_offload,			0, 0, 0, 0, 0, 0, NIX_RX_OFFLOAD_NONE)	\
+R(rss,				0, 0, 0, 0, 0, 1, RSS_F)		\
+R(ptype,			0, 0, 0, 0, 1, 0, PTYPE_F)		\
+R(ptype_rss,			0, 0, 0, 0, 1, 1, PTYPE_F | RSS_F)	\
+R(cksum,			0, 0, 0, 1, 0, 0, CKSUM_F)		\
+R(cksum_rss,			0, 0, 0, 1, 0, 1, CKSUM_F | RSS_F)	\
+R(cksum_ptype,			0, 0, 0, 1, 1, 0, CKSUM_F | PTYPE_F)	\
+R(cksum_ptype_rss,		0, 0, 0, 1, 1, 1, CKSUM_F | PTYPE_F | RSS_F)\
+R(vlan,				0, 0, 1, 0, 0, 0, RX_VLAN_F)		\
+R(vlan_rss,			0, 0, 1, 0, 0, 1, RX_VLAN_F | RSS_F)	\
+R(vlan_ptype,			0, 0, 1, 0, 1, 0, RX_VLAN_F | PTYPE_F)	\
+R(vlan_ptype_rss,		0, 0, 1, 0, 1, 1, RX_VLAN_F | PTYPE_F | RSS_F)\
+R(vlan_cksum,			0, 0, 1, 1, 0, 0, RX_VLAN_F | CKSUM_F)	\
+R(vlan_cksum_rss,		0, 0, 1, 1, 0, 1, RX_VLAN_F | CKSUM_F | RSS_F)\
+R(vlan_cksum_ptype,		0, 0, 1, 1, 1, 0,			\
+			RX_VLAN_F | CKSUM_F | PTYPE_F)			\
+R(vlan_cksum_ptype_rss,		0, 0, 1, 1, 1, 1,			\
+			RX_VLAN_F | CKSUM_F | PTYPE_F | RSS_F)		\
+R(mark,				0, 1, 0, 0, 0, 0, MARK_F)		\
+R(mark_rss,			0, 1, 0, 0, 0, 1, MARK_F | RSS_F)	\
+R(mark_ptype,			0, 1, 0, 0, 1, 0, MARK_F | PTYPE_F)	\
+R(mark_ptype_rss,		0, 1, 0, 0, 1, 1, MARK_F | PTYPE_F | RSS_F)\
+R(mark_cksum,			0, 1, 0, 1, 0, 0, MARK_F | CKSUM_F)	\
+R(mark_cksum_rss,		0, 1, 0, 1, 0, 1, MARK_F | CKSUM_F | RSS_F)\
+R(mark_cksum_ptype,		0, 1, 0, 1, 1, 0, MARK_F | CKSUM_F | PTYPE_F)\
+R(mark_cksum_ptype_rss,		0, 1, 0, 1, 1, 1,			\
+			MARK_F | CKSUM_F | PTYPE_F | RSS_F)		\
+R(mark_vlan,			0, 1, 1, 0, 0, 0, MARK_F | RX_VLAN_F)	\
+R(mark_vlan_rss,		0, 1, 1, 0, 0, 1, MARK_F | RX_VLAN_F | RSS_F)\
+R(mark_vlan_ptype,		0, 1, 1, 0, 1, 0,			\
+			MARK_F | RX_VLAN_F | PTYPE_F)			\
+R(mark_vlan_ptype_rss,		0, 1, 1, 0, 1, 1,			\
+			MARK_F | RX_VLAN_F | PTYPE_F | RSS_F)		\
+R(mark_vlan_cksum,		0, 1, 1, 1, 0, 0,			\
+			MARK_F | RX_VLAN_F | CKSUM_F)			\
+R(mark_vlan_cksum_rss,		0, 1, 1, 1, 0, 1,			\
+			MARK_F | RX_VLAN_F | CKSUM_F | RSS_F)		\
+R(mark_vlan_cksum_ptype,	0, 1, 1, 1, 1, 0,			\
+			MARK_F | RX_VLAN_F | CKSUM_F | PTYPE_F)		\
+R(mark_vlan_cksum_ptype_rss,	0, 1, 1, 1, 1, 1,			\
+			MARK_F | RX_VLAN_F | CKSUM_F | PTYPE_F | RSS_F)	\
+R(ts,				1, 0, 0, 0, 0, 0, TS_F)			\
+R(ts_rss,			1, 0, 0, 0, 0, 1, TS_F | RSS_F)		\
+R(ts_ptype,			1, 0, 0, 0, 1, 0, TS_F | PTYPE_F)	\
+R(ts_ptype_rss,			1, 0, 0, 0, 1, 1, TS_F | PTYPE_F | RSS_F)\
+R(ts_cksum,			1, 0, 0, 1, 0, 0, TS_F | CKSUM_F)	\
+R(ts_cksum_rss,			1, 0, 0, 1, 0, 1, TS_F | CKSUM_F | RSS_F)\
+R(ts_cksum_ptype,		1, 0, 0, 1, 1, 0, TS_F | CKSUM_F | PTYPE_F)\
+R(ts_cksum_ptype_rss,		1, 0, 0, 1, 1, 1,			\
+			TS_F | CKSUM_F | PTYPE_F | RSS_F)		\
+R(ts_vlan,			1, 0, 1, 0, 0, 0, TS_F | RX_VLAN_F)	\
+R(ts_vlan_rss,			1, 0, 1, 0, 0, 1, TS_F | RX_VLAN_F | RSS_F)\
+R(ts_vlan_ptype,		1, 0, 1, 0, 1, 0, TS_F | RX_VLAN_F | PTYPE_F)\
+R(ts_vlan_ptype_rss,		1, 0, 1, 0, 1, 1,			\
+			TS_F | RX_VLAN_F | PTYPE_F | RSS_F)		\
+R(ts_vlan_cksum,		1, 0, 1, 1, 0, 0,			\
+			TS_F | RX_VLAN_F | CKSUM_F)			\
+R(ts_vlan_cksum_rss,		1, 0, 1, 1, 0, 1,			\
+			MARK_F | RX_VLAN_F | CKSUM_F | RSS_F)		\
+R(ts_vlan_cksum_ptype,		1, 0, 1, 1, 1, 0,			\
+			TS_F | RX_VLAN_F | CKSUM_F | PTYPE_F)		\
+R(ts_vlan_cksum_ptype_rss,	1, 0, 1, 1, 1, 1,			\
+			TS_F | RX_VLAN_F | CKSUM_F | PTYPE_F | RSS_F)	\
+R(ts_mark,			1, 1, 0, 0, 0, 0, TS_F | MARK_F)	\
+R(ts_mark_rss,			1, 1, 0, 0, 0, 1, TS_F | MARK_F | RSS_F)\
+R(ts_mark_ptype,		1, 1, 0, 0, 1, 0, TS_F | MARK_F | PTYPE_F)\
+R(ts_mark_ptype_rss,		1, 1, 0, 0, 1, 1,			\
+			TS_F | MARK_F | PTYPE_F | RSS_F)		\
+R(ts_mark_cksum,		1, 1, 0, 1, 0, 0, TS_F | MARK_F | CKSUM_F)\
+R(ts_mark_cksum_rss,		1, 1, 0, 1, 0, 1,			\
+			TS_F | MARK_F | CKSUM_F | RSS_F)\
+R(ts_mark_cksum_ptype,		1, 1, 0, 1, 1, 0,			\
+			TS_F | MARK_F | CKSUM_F | PTYPE_F)		\
+R(ts_mark_cksum_ptype_rss,	1, 1, 0, 1, 1, 1,			\
+			TS_F | MARK_F | CKSUM_F | PTYPE_F | RSS_F)	\
+R(ts_mark_vlan,			1, 1, 1, 0, 0, 0, TS_F | MARK_F | RX_VLAN_F)\
+R(ts_mark_vlan_rss,		1, 1, 1, 0, 0, 1,			\
+			TS_F | MARK_F | RX_VLAN_F | RSS_F)\
+R(ts_mark_vlan_ptype,		1, 1, 1, 0, 1, 0,			\
+			TS_F | MARK_F | RX_VLAN_F | PTYPE_F)		\
+R(ts_mark_vlan_ptype_rss,	1, 1, 1, 0, 1, 1,			\
+			TS_F | MARK_F | RX_VLAN_F | PTYPE_F | RSS_F)	\
+R(ts_mark_vlan_cksum_ptype,	1, 1, 1, 1, 1, 0,			\
+			TS_F | MARK_F | RX_VLAN_F | CKSUM_F | PTYPE_F)	\
+R(ts_mark_vlan_cksum_ptype_rss,	1, 1, 1, 1, 1, 1,			\
+			TS_F | MARK_F | RX_VLAN_F | CKSUM_F | PTYPE_F | RSS_F)
+
 #endif /* __OTX2_RX_H__ */